
How do you sort e-commerce semantic search results by price, rating, best sellers, or freshness, while keeping a good precision?
https://docs.vespa.ai/en/reference/rank-features.html Not sorting, obviously. Visitors will never click on ill-matched products, even with the best ratings or the cheapest price. Without further ado, the correct answer is …. ???? ranking !!! Ranking is giving a score to results, based on expressions. Results are then sorted on this score. To get a good balance between matching and other internal or external attributes, the ranking has to moderate matching scores with other scores, in a mathematical expression. The problem is to be able to do this ranking with good performance. For all that reasons, Vespa.ai big data, three-phases ranking expressions, is an excellent solution: – First-phase and second-phase ranking expressions can be computed on all stateful content nodes – An addition third-phase (named global phase) can also be computed





